anvil-py 0.1.0

AI-powered agentic dependency manager that analyzes changelogs and safely upgrades packages.

Anvil - An Agentic Dependency Manager

Anvil is an AI-powered dependency upgrade system that autonomously analyzes, reasons about, and applies package updates.

Features

✧ Autonomous upgrade analysis with AI-powered reasoning
✧ Multi-format support: requirements.txt, poetry, uv, and conda environments
✧ Breaking change prediction and mitigation strategies
✧ Safe rollback on failure
✧ Changelog forensics to understand what's changing

Installation

Recommended: Install via pipx for isolation.

pipx install anvil-py

Or with pip (not recommended as it may conflict with project dependencies):

pip install anvil-py

Configuration

Anvil requires an LLM to perform AI analysis. You have two options:

Option 1: Local Ollama (Free)

Install and run Ollama locally:

# Install Ollama, then pull a model
ollama pull llama3.2

Anvil will automatically use Ollama if it's running on localhost:11434.

Option 2: OpenAI API

Set your OpenAI API key:

export OPENAI_API_KEY="sk-..."

Anvil will use OpenAI's models for analysis when the API key is present.

Quick Start

Go to your project directory (supports pip, poetry, or uv) and run:

anvil upgrade

Anvil will:

1. Detect your dependencies.
2. Check for updates.
3. Fetch changelogs & perform AI Forensic Analysis.
4. Safely Trial Install, Test, and Commit upgrades.

Usage

Interactive Upgrade (Recommended)

The main way to use Anvil is the interactive wizard:

anvil upgrade

Analysis Only (Dry Run)

If you just want to see what packages are outdated and read their changelogs without making changes:

anvil check .

Check Version

anvil version

Development

If you want to contribute to Anvil:

Check out TODO.md for a list of planned features and improvements you can help with! Other contributions are also welcome! Please feel free to submit a Pull Request. For major changes, please open an issue first to discuss what you would like to change.

# Run tests
pytest